Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.

Patents

  1. Advanced Patent Search
Publication numberUS6697885 B1
Publication typeGrant
Application numberUS 09/576,591
Publication dateFeb 24, 2004
Filing dateMay 22, 2000
Priority dateMay 22, 1999
Fee statusLapsed
Publication number09576591, 576591, US 6697885 B1, US 6697885B1, US-B1-6697885, US6697885 B1, US6697885B1
InventorsAnthony E. B. Goodfellow
Original AssigneeAnthony E. B. Goodfellow
Export CitationBiBTeX, EndNote, RefMan
External Links: USPTO, USPTO Assignment, Espacenet
Automated DMA engine for ATA control
US 6697885 B1
Abstract
An automated direct memory access system is implemented as an advanced ATA host IC for mother board or adapter applications. The system transfers data from two independent ATA channels using the ATA Ultra-100 protocols. The ADMA implements a command chaining technique to de-couple the host command sequence from channel execution. Software builds a command chain for hardware execution. The ADMA hardware independently reads command chain requests from memory and executes the next task on the list. When the ADMA hardware completes a task, it interrupts the host in order to inform the host that the task is complete, but immediately proceeds to the next task without waiting for interrupt servicing by the host.
Images(5)
Previous page
Next page
Claims(2)
What is claimed is:
1. In an ATA compatible data transfer system, an automated direct memory access (ADMA) system comprising:
a PCI interface portion for communicating with a host computer;
a BIOS interface portion adapted for communication with a flash memory device;
at least a first ATA interface for communicating with an ATA device; and
a set of ADMA internal registers, wherein the internal registers include command parameter block (CPB) information, disposed in a CPB chain defining a linked list of CPB entries, each CPB entry containing at least the ATA commands for an ATA device, and wherein the ADMA system controls an ATA channel in accordance with sequential linked commands without requiring interrupt servicing by the host computer between channel command executions.
2. In an ATA compatible data transfer system, an automated direct memory access (ADMA) method comprising:
providing a set of command parameter blocks (CPB), the blocks disposed in a CPB chain defining a linked list of CPB entries, each CPB entry containing at lest the ATA commands for an ATA device;
providing a set of physical region descriptors (PRD), a set of PRDs associated with each CPB, each PRD defining a memory location to which data is read from or written to; and
sequentially executing entries in a plurality of CPBs, without requiring interrupt servicing by the host computer between channel command executions.
Description

This application claims the benefit of provisional application Ser. No. 60/135,445 filed May 22, 1999.

FIELD OF THE INVENTION

The present invention relates to ATA-to-peripheral device bi-directional data transfer systems, and more particularly, to an advanced ATA adapter implementing a command chaining technique to de-couple the host command sequence from channel execution.

BACKGROUND OF THE INVENTION

The performance of ATA devices has increased dramatically over the last few years, particularly following introduction of the overlapped and queued command set. However, systems using these faster devices have not exhibited all of the expected benefits. This limited performance improvement can be traced to the design and implementation of the ATA host adapters currently in use, with the problem being amplified by the design of the Standard ATA driver software used in Windows® operating systems. These drivers effectively treat ATA transfers as a single-threaded entity, with correspondingly long latencies, required to have a host system fully service interrupt requests.

There are a number of factors that contribute to the time taken by a host computer to service a I/O request. In most operating systems, register access requires a subroutine call involving considerable processor overhead. The register access sequence of actions requires seven separate PCI 8-bit I/O transfers. Each register access takes a minimum of 4 PCI clocks (120 ms at 33 MHz) plus the ATA cycle time for each transfer. In many cases, the ATA devices are set to mode 0 for command-block transfers. In mode 0, the ATA cycle time is 600 ms for each access. This time is in addition to the 120 ns required for a PCI transfer, making a total of 720 ns per register access or 5040 ns (5 us) for a complete command set up sequence. During this period, the PCI bus and the processor are unavailable for further functions.

The data transfer period depends on the ATA protocol in use, the size of the ATA adapter FIFO, the size of the host cache line, and the overall PCI bus utilization. During bus master transfers, the memory and PCI usage is interleaved with the processor's requirements. Thus, the processor is able to fetch instructions and process them within its own pipeline, concurrent with the bus-master transfer. The normal unit of transfer requested by a Microsoft O/S is 4k Bytes. For an Ultra 100 system operating on a PCI-33 bus, the minimum transfer time would be 30 us, making the command set up time a minimum of 16% of the transfer time. When the transfer is complete, the ATA device may assert an interrupt. The host shall then read the ATA status register to determine if the transfer was successful and to clear any pending interrupt. Thus, for any one transfer, there is the potential for a long period of PCI bus utilization for command transfer and the need to service an interrupt.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a semi-schematic block level diagram of an ADMA architecture in accordance with the invention;

FIG. 2 is a semi-schematic block level diagram of an exemplary Windows® driver;

FIG. 3 is a semi-schematic block level diagram of the operational components of an ADMA system in accordance with the invention;

FIG. 4 is a semi-schematic block level flow diagram of ADMA memory usage illustrating the CPB chain.

2 Introduction

Although the performance of ATA Devices has increased drastically over the last few years, systems using them have not shown all of the corresponding benefits. This limited performance improvement can be traced to the design and implementation of the ATA host adapters currently in use. The problem in Wintel systems is amplified by the design of the standard ATA driver used in Windows® operating systems. These drivers effectively treat ATA transfers as a single-threaded entity, with correspondingly long latencies to fully service interrupts. The ADMA is designed to address these problems and add features that will make it and ATA Devices more suitable for true multi-threading applications, such as Network server and streaming-AV applications.

The objective of the ADMA design is to drastically increase the performance of systems that use ATA Devices.

3 Normative References

The following documents contain terminology definitions, protocol descriptions, and implementation standards which must be taken into account when discussing the content of this specification. All standards are subject to revision, and parties to agreements based on this standard are encouraged to investigate the possibility of applying the most recent editions of the standards listed below, and, where applicable, tracking the draft versions of forthcoming revisions to those standards.

3.1 Industry Standard References

3.1.1 ANSI Information Technology—AT Attachment with Packet Interface-5 (ATA/ATAPI)

Published standard: AT Attachment Interface with Packet Interface Extensions (ATA/ATAPI-4), ANSI National Standard for Information Systems, #X3.317-1998

Draft standard: AT Attachment Interface with Packet Interface Extensions (ATA/ATAPI-5), Proposed Draft #T13-1321D Revision 1, ANSI National Standard for Information Systems, #X3.317-1998

3.1.2 PCI Local Bus Specification—Revision 2.2

Standard for the PCI Local Bus, published Dec. 18, 1998, by the PCI Special Interest Group (PCI SIG)

3.2 GSI Development Specifications

3.2.1 GSI Product Specification: HotPort™ External Device Interface

GSI Product Specification 1646-PS-02-6. The HotPort™ External Device Interface

3.2.2 GSI Technology Specification: ATA/ATAPI Sub-Channel Implementation

GSI Technology Specification 1843-PS-03-6

4 Definitions, Abbreviations, and Conventions

4.1 Definitions and Abbreviations

4.1.1 ATA Adapter

The hardware that performs the interface between the ATA Host and the ATA channel. The embodiment can be an IC or plug-in adapter.

4.1.2 ATA Bus

The physical connection between an ATA Adapter and an ATA Device, consisting of conductors carrying signals.

4.1.3 ATA Channel

The ATA channel is the logical transport mechanism between the ATA host and the ATA Devices on an ATA Bus. Each channel can have two devices connected to it.

4.1.4 ATA sub-channel

A sub-channel controller can replace an ATA Device. A sub-channel controller can host further ATA channels allowing extra devices to be addressed.

4.1.5 ATA Device {ATA Spec.}

A data storage peripheral. Traditionally, a Device on the ATA interface has been a hard drive, but any form of storage device may be placed on the interface, provided that the Device adheres to this ATA standard.

4.1.6 ATA Host

Software is executed ion the host that controls the functions of the ATA-subsystem.

4.1.7 ATA subsystem

The hardware elements that include an ATA Adapter, ATA channel, and the ATA Device(s).

4.1.8 ATA-n

A shorthand reference to the standard specified in the ATA-n (or ATA/ATAPI-n, as applicable) standards document, whether published as final, circulated in draft form, or only in the planning stage.

4.1.9 ATA Standard

The current, evolving ATA/ATAPI standard. The standard specified by the draft of the ATA/ATAPI interface specification, NCITS document T13 1321D Revision xx, current at the time of the writing of this document.

4.1.10 ATA Specification

The draft of the ATA/ATAPI interface specification, NCITS document T13 1321D Revision xx current at the time of the writing of this document.

4.1.11 ATAPI (AT Attachment Packet Interface) Device

A Device which implements the Packet Command feature set.

4.1.12 bus protocol

The sequence of bus signal states, and their timings, which are required in order to transfer commands and data along a bus. There may be more than one protocol available on any one bus.

4.1.13 ATA bus release {ATA Spec.}

The act of clearing both DRQ and BSY to zero before the action requested by the command is completed, to allow the Host to select the other Device on the channel. (Applies only to Devices that implement overlap.).

4.1.14 command queue

The set of all commands for which command acceptance has occurred but command completion has not occurred. Command queuing allows the Host to issue concurrent commands to the same Device. Only commands in the Overlapped Feature Set may be queued.

4.1.15 CRC (Cyclic Redundancy Check) {ATA Spec.}

Used for the Ultra DMA protocol to check the validity of the data that has been transferred during the last Ultra DMA burst.

4.1.16 DMA (Direct Memory Access) {ATA Spec.}

A means of data transfer, between Device and Host memory, without Host processor intervention.

4.1.17 LBA (Logical Block Address) {ATA Spec.}

A mode of addressing an ATA Device according to a linear mapping of sectors.

4.1.18 Overlap {ATA Spec.}

A protocol that allows Devices that require extended command execution time to perform an ATA bus release, so that commands may be executed by another Device on the same bus.

4.1.19 PIO (Programmed Input/Output) {ATA Spec.}

A means of accessing Device registers. The term PIO is also used to describe one form of data transfers. PIO data transfers are performed by the host processor, utilizing PIO accesses to the Data Register.

4.1.20 Ultra DMA Burst{ATA Spec.}

The period from an assertion of DMACK—to the subsequent negation of DMACK—when Ultra DMA has been enabled by the Host.

4 1.21 Wintel

A shorthand reference to systems which use Intel X86 and MS Windows® architecture.

4.2 Conventions

4.2.1 Names of Registers, Words, Bytes, Bits, etc.

The names of registers, words, bytes, bits, and modes begin with uppercase letters.

In addition, register names are prefixed with the acronyms PCI, ADMA, or ATA, to indicate which of these register sets they belong to.

For example: ATA Status Register, Status Word, Bytes 0-3, Error Bit.

4.2.2 Signal Names and Signal States

Signal names are shown in all uppercase letters. All signals are either active-high or active-low signals. If a hyphen (“-”) character is affixed to the end of a signal name, it indicates that the signal is an active-low signal; otherwise, the signal is active-high. An active-low signal is true when the signal is below VIL, and it is false when the signal is above VIH. An active-high signal is true when the signal is above VIH, and it is false when the signal is below VIL.

A signal is ‘asserted’ when it is driven by an active circuit to the true state. A signal is ‘negated’ when it is driven by an active circuit to the false state. A signal is ‘released’ when it is not actively driven to any state. Some signals have bias circuitry that pulls the signal to either a true state or a false state when no signal driver is actively asserting or negating the signal.

4.2.3 Numbers

Numbers are decimal, unless specified otherwise.

Hexadecimal numbers are shown as a string of digits, 0 through 9 or A through F, followed by ‘h’: e.g. ‘1AB7h.’

Binary numbers are shown as a string of digits, 0 or 1, followed by ‘b’: e.g. ‘10110111 b.’

4.2.4 Content Imported from Normative Specifications

Where material within this specification has been imported from another, normative, specification, it is done so for the sake of self-sufficiency of this document and the normative source is identified. (For exact references to the normative specifications cited here, see the Normative Documents section).

For example:

{PCI Spec} indicates that this item is imported from the PCI Local Bus Specification version which was current at the time when this document was written.

{ATA Spec} indicates that this item is imported from the ATA/ATAPI Specification version which was current at the time when this document was written.

5 Overview

The ADMA is an advanced ATA adapter IC for motherboard or adapter applications. It transfer data from two ATA channels using the ATA Ultra-100 protocols on PCI66 MHz bus. To fully optimize the systems throughput it implements a command chaining technique to de-couple the host command sequence from the channel execution, allowing true multitasking. The device is implemented in 0.35 micron CMOS technology and available in a 176-pin package.

5.1 Summary of Features

Legacy Compatible—Works with existing drivers.

Bus Master Transfers to and from Memory—Minimizes CPU load.

Bus Master Transfers to and from I/O—Makes direct ATA-to-peripheral transfers possible (e.g., AV transfers from disk to MPEG2 adapters without system memory usage).

Command Chaining—Maximizes system throughput, by allowing next-command execution to start while a previous-command completion interrupt is being serviced by the Host.

HotPort™ Interface—Allows low-cost connection to externally mounted ATA Devices and adds sub-channel capability.

Unsolicited Interrupt Support—Provides insertion/removal request notification for removable Devices.

Enhanced Data Integrity—Allows long-datablock transfers without degrading Ultra ATA CRC effectiveness.

PCI 2.2 Hot Plug Compliance—Allows ADMA use in Hot Plug systems.

PCI 2.2 Low-Power Compliance—Supports system low-power modes.

ATA66/100 Electrical Interface Drivers—Meets ATA for slew-control specifications over the entire specified electrical and temperature range. Reduces noise and ringing effects.

Ultra ATA 100—Supports planned Ultra ATA speedup, up to 100 MB/s per channel.

PCI 66—Operates at 66 MHz on the PCI bus.

True Dual-Channel Design—Provides fully independent channels, with no multiplexing of any pins, eliminating associated stray capacitance and noise problems.

Large (128k-Byte) Flash BIOS Support—Suitable for adapters designed for a wide variety of PCI Bus platforms, including Wintel, Apple, AS400, and others

4k-Byte Read Write FIFOs—Maximizes data throughput.

Auto Polling of IRQs—Allows efficient implementation of overlapped commands.

Up to 256 Sub-Channels per ATA Channel—Allows 512 Devices on a single ATA channel.

ADMA register reporting of PCI error conditions—Easier Driver response to exception conditions.

Large LBA Proposal Compatability—Compatible with current proposal or methods of addressing ATA Devices with capacities greater than 138GB.

JTAG—Chip and board level testing.

6 Host Overhead Reduction

There are a number of factors that contribute to the time taken by the Host to service an I/O request. This section examines the requirements of the ATA protocol and the design of Windows™ drivers, both of which contribute to the time taken to service an I/O request. ADMA command chaining is shown to drastically reduce the overhead.

6.1 ATA Protocol Requirements

To set up a command on an ATA Device, the Host will typically perform the following sequence of events:

1—Read the ATA Status Register to determine that the channel is not busy.

2—Write the ATA Device Head Register to ensure that the correct Device is selected.

3—Write the ATA Sector Count Register to indicate the number of sectors to be transferred.

4—Write the ATA Sector Number Register to load part of the LBA address.

5—Write the ATA Cylinder Low Register to load part of the LBA address.

6—Write the ATA Cylinder High Register to load part of the LBA address.

7—Write the ATA Command Register with the command to be executed.

This sequence of actions requires 7 separate PCI 8-bit I/O transfers. If the ATA Adapter is able to internally buffer these commands, the process takes a minimum of 4 PCI cycles (840 ns at 33 MHz). During this time, all processor activity and PCI activity is stopped. If the ATA adapter cannot buffer these I/O writes, the PCI bus has to wait the ATA cycle time for each transfer. In many cases, the ATA Devices are set to Mode 0 for command-block transfers. In Mode 0, the ATA cycle time is 600 ns for each access. This time is in addition to the 840 ns required for the PCI transfer, making a minimum total of 5040 ns (5 us) during which period the PCI bus and the processor are unavailable for other functions.

The data-transfer period depends on the ATA protocol in use, the size of the ATA Adapter FIFO, the size of the Host cache-line, and the overall PCI bus utilization. These factors are discussed in 7. During bus master transfers, the memory and PCI usage is interleaved with the processor's requirements. Thus the processor is able to fetch instructions and process them within its own pipeline, concurrent with a bus-master transfer. The normal unit of transfer requested by the O/S is 4k Bytes. For an Ultra 100 system operating on a PCI-66 bus, the minimum transfer time would be 40 us, making the command setup time 25% of the transfer time.

When the transfer is complete, the ATA Device will assert an interrupt. The Host then must read the ATA Status Register to determine if the transfer was successful and to clear the pending interrupt. The treatment of interrupts by the Host software driver is discussed in Section 6.2.

Thus for any one transfer there is the potential for a long period of PCI bus utilization for command transfer and the need to service an interrupt.

6.2 Windows® Driver Structure

The servicing of interrupts on Wintel systems has a dominating effect on ATA subsystem performance. The reasons for this can be explained by examining the requirements of both the ATA protocol and the Windows® driver requirements.

FIG. 1 is a simplified illustrator of the Windows® driver structure. An application makes a request for data from the I/O Manager. The IOM then builds the request into an I/O Request Parameter block (IRP), and the IRP is passed by the IOM to various elements required to service the request. The request will be passed to the File system whose function is to determine where the data is located and add additional information request to the IRP.

The IRP is then passed back to the data manager with a request to pass it on to the driver for the Device. When the driver receives the IRP it transforms the request into a form that can be understood and executed by the hardware.

The request is processed by the driver in four phases: Dispatch, Start I/O, IRQ, and Completion.

6.2.1 Dispatch Phase

The IRP is accepted and queued.

6.2.2 Start I/O

The Driver passes the formatted information to the hardware. The IRP is updated to show that an interrupt is expected and control is passed back to the IOM.

6.2.3 IRQ

When the ATA Device asserts an interrupt the IOM has to determine where the interrupt originated from. On PCI systems, many Devices can use the same interrupt signal. The IOM has to determine which Device caused the interrupt, to do this it enters the IRQ Phase, each driver is called in turn to determine to if it has control of the interrupting Device. In this phase, the driver's only allowed actions are to reset the interrupt, store the status and return with a request indicating that it needs to be re-entered in the Completion Phase.

6.2.4 Completion Phase

During this phase the driver checks the status conditions and updates the IRP.

In standard ATA drivers the IOM can only pass another request to the driver after finishing the Completion Phase. On real systems, it can be hundreds of microseconds before the IOM re-enters the driver in the Completion Phase. If the IOM calls the driver with another IRP before the current IRP has completed its Completion Phase, the driver will reject the IRP informing the IOM that it must queue the IRP until the driver is free.

The actions of the software driver and of the hardware are closely coupled. Experience shows that the IRQ and Completion Phases can take more than 100 us to complete.

6.3 Command Chain Solution

The principle objective of command chaining is to allow the software driver and the hardware to be loosely coupled. To do this, the software can build up a list of tasks (command chain) for the hardware to complete. The hardware independently reads these requests from memory and executes the task. When the hardware completes a task, it will interrupt the Host to inform the Host that the task is complete, but will immediately proceed to the next task without waiting.

Thus, in the case of a Windows™ driver, the IOM will receive an IRP and put it onto the command chain. It will then return to the DM. If the IOM passes-in another IRP, it too will be put onto the chain. IRP requests can be added onto the chain until the driver's workspace is exhausted. Concurrent with Host request setup activity, the hardware is reading the command chain and executing requests. After each request is complete, the hardware asserts an IRQ and the IRQ Phase is entered. The driver then requests a Completion Phase for the appropriate IRPs. The ADMA does not have to wait for the IRP or the Completion Phase but can immediately continue with the next task in the chain.

6.3.1 Command Chain Performance Benefits

The overhead associated with writing the command to the Device drops from 5 us, 25% of the transfer time (Section 6.1), to 4 PCI clocks (60 ns) at PCI-66, or 0.15% of the transfer time.

The delay imposed between execution of one I/O request and initiation of the next drops from a typical 100 us to approximately 360 ns (24 PCI clocks at PCI-66).

7 Major Parameters

The features of the ADMA are designed to provide full legacy compatibility along with many new advances to drastically improve performance as well as functionality. This section describes the features and the reasons why they have been specified.

7.1 FIFO Size

7.1.1 Conclusion

All current Ultra ATA Host controller implementations use a FIFO to buffer the data between the Device and the Host. It has been observed that these FIFOs are insufficient in size to allow the continuous transfer of data. A FIFO size of 4k Bytes (plus 128 Bytes for over-run) in each direction [rough!] has been chosen for the ADMA.

The following factors have been taken into consideration when determining the size of the FIFOs used in the ADMA.

7.1.2 PCI Bus Latency

To perform a Bus Master transfer, the ADMA requests control of the PCI Bus. When this grant is requested, it can transfer data until the grant is removed or no more data is available. The time the Host takes to grant a bus control request varies, depending on the utilization of the PCI bus. This time can vary from 11 to several hundred bus clocks. Thus when the grant is given, the ADMA should have sufficient data or space available in its FIFO to be able to continuously transfer data until the grant is removed. This reduces the number of request/grant sequences required, consequently reducing bus occupancy and reducing the overall transfer time.

7.1.3 Ultra ATA Pausing

If the FIFO becomes empty or full due to the effects of PCI bus latency, the ADMA will pause the Ultra DMA transfer. Such pauses dramatically reduce the overall transfer rate and increase the transfer time. A large FIFO will reduce the likelihood of this occurrence.

7.1.4 PCI Bus Speed

The PCI bus commonly used in Wintel systems is 32-bit/33 MHz. This bus can theoretically transfer data at 133 MB/s. This speed is attained only in short bursts. Sustained bursts of 80 to 100 MB/s can be expected on most systems. It should be possible to sustain transfers from a single Ultra-66 channel. but dual Ultra-66 transfers cannot be sustained. Thus a FIFO capable of holding an average transfer will result in the least number of Ultra pauses. Single-channel transfer from Ultra-100 may be sustainable on some systems.

7.1.5 Operating System Transfer Size

The usual unit of transfer used by Windows ® Operating Systems is 4k (4096) Bytes.

7.2 PCI Bus Speed and Width

7.2.1 Conclusion

The ADMA will be a 32-bit, 66 MHz PCI device. The following factors have been taken into consideration when coming to this conclusion.

7.2.2 System requirements

Currently (1999), most systems are shipping with 33 MHz, 32-bit bus solutions. It is anticipated that, by the end of 1999, 66 MHz systems will start to become predominant in high-end systems. However, 64-bit-wide operation is not seen as being a factor for another 12 to 24 months.

7.2.3 Bus Width

As explained in Section 7.2.2, 32-bit operation is the predominant system implementation. 64-bit operation would increase performance and decrease PCI bus utilization. However, 64-bit operation would increase the external pin count, and thus cost, for no real system benefits. An increase to 64-bit operation could be an enhancement for future revisions of the ADMA device.

7.2.4 Ultra Performance

Section 7.1.4 outlines the effects of PCI performance on the ability to service the Ultra ATA devices. 33 MHz PCI buses can only maintain performance on a single Ultra ATA channel. For Ultra 100 and two-channel Ultra 66 operation, 66 MHz operation is required for optimum performance. The system will operate reliably with a 33 MHz bus.

7.3 PCI Specification

7.3.1 Conclusion

The ADMA device will conform to PCI 2.2, operating at 3.3 V or 5 V, and will be Hot Plug capable.

7.3.2 Low-Power Mode

The ADMA supports the pCLKRUN signal. When in the idle state the ADMA will release the pCLKRUN signal, this will allow the host to reduce the PCI clock to 0 Hz so reducing the power consumption of those elements of the core operating on the PCI clock. To further reduce power the ADMA supports the PCI low-power state “D2”. In this state the internal clock is stopped but the register contents are maintained. Upon wake up the transition from the “D2” state to the operational “D0” state provides the time needed for the internal PLL to stabilize. Further power reductions require the removal of power to the ATA devices and thus a full initialization process. It should be noted that where this device is used on HotPort™ implementations the HotPort™ interface includes a signal that is intended to bring a system out of the “D3” state, this signal is not however passed through the ADMA.

7.3.3 Hot Plugging

A target market for the ADMA device is the low-cost server market. The device will be implemented on plug-in adapter cards. Thus the Hot Plug Specification will be followed.

7.3.4 Operating Voltage

Increasingly, PCI-based systems are using 3.3 V and 5 V signaling. It is anticipated that some systems will become 3.3 V-only. Thus the ADMA should be capable of being built into 3.3 V or 5 V systems and used on dual-mode PCI adapter cards.

The device will be a 3.3 V part with 5 V tolerance. Operation at 3.3V complies with the signal levels for ATA-4 and ATA-5.

7.4 ATA Protocol Support

7.4.1 Conclusion

The ADMA will operate in Legacy PIO Mode, DMA-Assisted PIO Mode, and Full DMA Mode.

7.4.2 ATA v Host Protocols

It is easy to confuse the terms describing the transfer protocols used in a PC system. The ADMA sits between the Host-system PCI Bus and the ATA Bus. The ADMA transfers data to and from the Host, using PCI protocols, and to and from ATA Devices, using ATA protocols. The terms PIO and DMA have meanings for both the ATA and PCI busses; however, they are not necessarily the same meaning. The protocols used on the PCI and ATA buses do not necessarily have to be the same for any one transfer.

7.4.3 PIO

For the Host, PIO is taken to mean that data is transferred between the Host and the ATA adapter by the Host reading or writing a Register in the ATA adapter using IN and OUT processor instructions. [rough!]

For the ATA bus, PIO means that data is transferred between the ATA Device and the ATA adapter by a sequence of signals on the ATA bus. The ATA adapter controls the sequence.

Host PCI-PIO is not an efficient method of data transfer. Host PIO is the legacy mode used on Wintel systems. Operation in this mode is defined as Legacy PIO.

The ATA standard defines various PIO transfer rates, the fastest of which is 16 MB/s. The ATA PIO protocol can be automated in the ADMA so that it can be used to transfer data to/from the Device while using PCI DMA from/to the Host. This method of operation is termed DMA-Assisted PIO Mode.

7.4.4 DMA

For the Host, DMA is taken to mean that data is transferred between the Host and the ATA adapter over the PCI bus, using the PCI Burst Mode protocol. Once initiated, the transfer requires no Host involvement. PCI Burst Mode DMA is the fastest method of data transfer over the ATA bus.

For the ATA bus, DMA means that data is transferred between the ATA Device and the ATA adapter by a sequence of signals on the ATA bus. The ATA adapter controls the sequence but is a different sequence from that used by ATA PIO. There are two methods of DMA defined: Multiword DMA and Ultra DMA.

7.4.4.1 ATA Multiword DMA

ATA Multiword DMA is defined to transfer data at up to 16 MB/s. This protocol has traditionally been used in conjunction with the PCI DMA protocol to provide a more efficient means of transferring data through the system. It seems that there are implementation imperfections in some ATAPI Devices. Consequently, most ATAPI host software programs use the PIO protocol.

The performance of Multiword DMA is insufficient for modern hard drives. It is likely that this protocol will be dropped in ATA-6.

7.4.4.2 Ultra-DMA

ATA Ultra-DMA is a high-speed mode of data transfer. ATA-5 defines transfer rates up to 66 MB/s, with ATA-6 slated to include a 100 MB/s rate. 32-bit ATA could increase this rate to 200 MB/s. In addition to high performance, the Ultra protocol defines the use of a CRC to validate that data has been correctly transferred.

ATA Ultra-DMA is used on all modern hard drives and is now starting to be implemented on ATAPI Devices.

7.4.5 DMA Method Implemented

Because DMA-Assisted PIO Mode equals the performance of Multiword DMA and does not have the same implementation problems on ATAPI Devices, the ADMA on supports Ultra DMA.[????] When using the ATA Ultra DMA protocol, Host transfers will be via PCI DMA. This mode is termed DMA Mode.

7.5 Operating Modes

7.5.1 Conclusion

Need three modes of operation: Legacy (power-on default), and SFF (for backward compatibility) and Command Chain Mode (for performance).

7.5.2 Legacy Modes

The basic operation of any ATA subsystem is in legacy PIO Mode. This means that the ATA adapter acts as an address decoder for the Host. All reads and writes are performed using Host I/O instructions. The only function performed by the ATA adapter is to control the signal timings of the ATA bus and to respond to the PCI signals. This mode of operation is compatible with standard BIOS code and all operating system drivers.

7.5.3 SFF Mode

The advent of PCI DMA ATA adapters introduced some new registers, features and conventions. Conforming to these legacy standards ensures compatibility with in-built operating system drivers that are DMA-aware. These registers and features are described in the SFF 8038i document. This document does not describe how the PCI adapter is to be configured; configuration is vendor specific.

7.5.4 Automatic DMA Mode

Section 6.3 describes the benefits of reduced overhead and increased throughput associated with command chains. Section 12 describes the operation in more detail.

8 ADMA Physical Definition

8.1 Pin and Package Definition

Package Type: 176-pin

Sides 1, 2, 3 & 4 progress counterclockwise around the package (as viewed from above). Within a side, the signals must be in the order shown (increasing pin numbers). Additional pins for power and ground may be inserted anywhere.

Note:

PCI signal names start with ‘p’

ATA channel X signal names start with ‘x’

ATA channel Y signal names start with ‘y’

Hotport™ signal names start with ‘xH’ or ‘yH’

Flash EPROM signal names start with ‘m’

I=input

O=output

O=output with tristate control

B=bidirectional

C=clock input

TABLE 1
Signal Names
Side Side Side Side
1-44 2-44 3-44 4-44
pins pins pins pins
PGNTn I pAD8 B yHRTDHn O
pREQn o pCBE0n B yHRTHDn I
pAD31 B pAD7 B xLEDn O yCS1n O
pAD30 B pAD6 B xCS1n O yCS0n O
pAD29 B pAD5 B xCS0 O yA2 O
pAD28 B pAD4 B xA2 O yA0 O
pAD27 B pAD3 B xA0 O yCBLID O
pAD26 B pAD2 B xCBLID I yA1 O
pAD25 B pAD1 B xA1 O yIRQ I
pAD24 B pAD0 B xIRQ I yACKn O
pCBE3n B mA6 O XACKn O yIORY I
pIDSEI B mA5 O xIORY I yIORn O
pAD23 I mA7 O xIORn O yIOWn O
pAD22 B mA4 O xIOWn O yDRQ I
pAD21 B mA12 O xDRQ I yUIRQn I
pAD20 B mA3 O xUIRQn I yD15 B
pAD19 B mA15 O xD15 B yD0 B
pAD18 B mAa2 O xD0 B yD14 B
pAD17 B mA16 O xD14 B yD1 B
pAD16 B mA1 O xD1 B yD13 B
pCBE2n B mWEN O xD13 B yD2 B
PFRAMEn B mA0 O xD2 B yD12 B
pIRDYn B mA14 O xD12 B yD3 B
pTRDYn B mD0 B xD3 B yD11 B
pDEVSELn B mA13 O xD11 B yD4 B
pSTOPn B mD1 B xD4 B yD10 B
pPERRn B mA8 O xD10 B yD5 B
pSERRn I mD2 B xD5 B yD9 B
pPAR B mA9 O xD9 B yD6 B
pCBE1n B mD3 B xD6 B yD8 B
pAD15 B mA11 O xD8 B yD7 B
pAD14 B mD4 B xD7 B yHHPRES O
pAD13 B mOEN O xRSTn O yRSTn O
pAD12 B mD5 B XTLI C pCLKRUNn O
pAD11 B mA10 O XTLO C pM66EN O
pAD10 B mD6 B yLEDn O pCLK C
pAD9 B mCEN O YHDPRES O pRSTn I
mD7 B yHDIRn I

8.1.1 Device Signal Definitions

8.1.1.1 pGNTn

Grant {PCI Spec} Indicates to the agent that access to the bus has been granted. This is a point-to-point signal. Every master has its own pGNTn, which must be ignored while pRSTn is asserted.

8.1.1.2 pREQn

Request {PCI Spec} Indicates to the arbiter that this agent desires use of the PCI bus. This is a point-to-point signal. Every master has its own pREQn, which must be tri-stated while pRSTn is asserted.

8.1.1.3 pAD[31::00]

Address and Data {PCI Spec} Address and data are multiplexed on the same PCI pins. A bus transaction consists of an address phase followed by one or more data phases. PCI supports both read and write bursts.

8.1.1.4 pCBE[3::0]n

Bus Command and Byte Enables {PCI Spec}

8.1.1.5 pIDSEL

Initialization Device Select {PCI Spec} Used as a chip select during configuration read and write transactions.

8.1.1.6 pFRAMEn

Cycle Frame {PCI Spec} Driven by the current master to indicate the beginning and duration of an access. pFRAMEn is asserted to indicate that a bus transaction is beginning. While pFRAMEn is asserted, data transfers continue. When pFRAMEn is de-asserted, the transaction is in the final data phase or has completed.

8.1.1.7 pIRDYn

Initiator Ready {PCI Spec}

8.1.1.8 pTRDYn

Target Ready {PCI Spec}

8.1.1.9 pDEVSELn

Device Select {PCI Spec}

8.1.1.10 pSTOPn

Stop {PCI Spec}

8.1.1.11 pPERRn

Parity Error {PCI Spec}

8.1.1.12 pSERRn

System Error {PCI Spec}

8.1.1.13 pPAR

Parity {PCI Spec}

8.1.1.14 mA[16::0]

8.1.1.15 mWEN

8.1.1.16 mD[7::0]

8.1.1.17 mOEN

8.1.1.18 mCEN

8.1.1.19 x/yLEDn

8.1.1.20 x/yHDPRES

8.1.1.21 x/yHDIRn

8.1.1.22 x/yHRTDHn

8.1.1.23 x/yHRTHDn

8.1.1.24 x/yCS1n

8.1.1.25 x/yCS0n

8.1.1.26 x/yA[2:0]

8.1.1.27 x/yIRQ

8.1.1.28 x/yACKn

8.1.1.29 x/yIORY

8.1.1.30 x/yIORn

8.1.1.31 x/yIOWn

8.1.1.32 x/yDRQ

8.1.1.33 x/yUIRQn

8.1.1.34 x/yD[15::00]

8.1.1.35 x/yHHPRES

8.1.1.36 x/yRSTn

8.1.1.37 pCLK

Clock {PCI Spec} Provides timing for all transactions on PCI and is an input to every PCI device. All other PCI signals, except RSTn, INTAn, INTCn, and INTDn, are sampled on the rising edge of CLK and all other timing parameters are defined with respect to this edge.

8.1.1.38 pRSTn

Reset {PCI Spec} Used to bring PCI-specific registers, sequencers, and signals to a constant state.

8.2 Electrical Characteristics

8.3 Environmental Constraints

Operating Temperature: 0° to 70° C. (ADMA chip case temperature)

9 Operation

9.1 ATA Protocol Support

The ADMA will operate in Legacy PIO Mode, DMA-Assisted PIO Mode, and Full DMA Mode. These modes are described in Section 7.4.

9.2 Modes of Operation

Each channel can operate in one of 3 modes defined by the setting of MODE in the ADMA Channel Register as written by the device driver: Legacy PIO Mode, DMA-Assisted PIO Mode, or Full DMA Mode

9.3 PIO Legacy Mode.

The Auto PIO timing generator has one set of timings for 8-bit ATA Command Register transfers and two for 16-bit ATA Data Register transfers (one for each Device on the cable). ATA Command Register timings reflect the slower Device on the cable and can be Modes 0-4. Data timings can be Modes 0-4.

PIO reads are performed within the PCI cycle, so that TRDYn is not asserted until the PIO data is available. PIO writes can assert TRDYn immediately provided the previous PIO write is complete.

9.4 SFF Legacy Mode

The Auto PIO timing generator is the same as in PIO Legacy Mode. ATA Command Register transfers are the same as in PIO Legacy Mode. ATA Data transfers pass through one of the FIFOs, thereby decoupling the ATA cycles from the PCI cycles. Timing is controlled either by the Ultra DMA timing generator (Modes 0-5) or the Auto PIO timing generator.

PRD table entries are read from memory into internal counting registers, using PCI Master Mode. The entries define the current transfer address and length. There can be more than one entry reflecting fragmentation of physical memory by the x86 paging registers.

PCI Master Mode is used to transfer bursts of data (controlled by the current PRD table entry) to or from memory. Synchronization is controlled by FIFO control signals (empty, full etc). In Auto PIO a Read Ahead Register is necessary to prevent the host from reading too much data, as the length counter in the current PRD entry is counting PCI transfers, not ATA transfers. In UltraDMA, the disk signals the end of the transfer by de-asserting DMARQ.

In Auto PIO, intermediate interrupts between blocks of data are not passed to the PCI INTAn.

PCI Master Mode is also used to transfer PRD table entries into the internal registers.

The PRD table structure, PRD Address Register, DMA Command Register, and DMA Status Register are described in Section( ).

9.5 Auto DMA Mode

Auto DMA Mode enables chains of ATA transfers to be defined in memory and the controller to execute them without assistance from the device driver. There will typically still be one interrupt per transfer to indicate completion, but the hardware does not have to wait for it to be serviced before starting the next transfer.

The Command Parameter Block Table is a singly linked circular table, with each entry containing of a list of ATA registers and the data to be written to them and whether they should be read at transfer completion. Each entry also contains a pointer to the PRD table for this transfer and a pointer to the next table entry.

The device driver formats the table, writes the start address into the CPB Address Register, and signals the hardware to start.

The hardware does a PCI Master Mode burst to load the PRD Address Register, write the ATA registers with the given data, and start the transfer. The sequence of events within the transfer is exactly the same as SFF Legacy Mode (unless Command Overlap and Queuing are enabled). At the completion of the transfer, the specified ATA registers are read and the status is updated in the memory image of the current CPB entry. A PCI INTAn is generated and, if no errors occurred, the next CPB table entry is read and executed.

The Command Parameter Block Table can contain entries for each Device on the cable, with the order of execution defined by the device driver.

9.6 Command Overlap and Queueing

If the Devices support Command Overlap and Queuing, the sequence becomes more complicated. Only certain commands may be overlapped or queued. On each such command a 5-bit tag value is written into one of the ATA registers. The command may complete as above or the Device may interrupt with status.

“Released”. In this case, the hardware proceeds to the next CPB entry (which may be for the other Device) and executes it. No PCI INTAn is generated.

In this way, queues of commands can be written to one or both Devices.

If overlapped commands are written to both Devices, the CPB table should include a Channel Command to initiate device polling. When a Device is ready to transfer data for any one of the queued commands, it interrupts with status “Service”. In this case the hardware searches for the CPB table entry with the matching tag, sets up the registers to define PRD Address from the CPB entry, and issues a “Service” command to start the transfer.

No PCI INTAn is generated.

10 PCI Registers

10.1 PCI Configuration Header Registers

The controller implements a subset of the standard type 00h configuration header register set. All registers have the standard meaning as defined in the PCI specification issue 2.1. The registers actually implemented and device-specific values are described below. Fields marked reserved are all zeros and read only.

TABLE 2
PCI Configuration Registers
Byte Bits Bits Bits Bits
Offset 31-24 23-16 15-8 7-0
00 h Device ID Vendor ID
04 h Status Command
08 h Class Code Revision
0 ch Reserved Header Type Latency Timer Cache Line
Size
10 h Base Address 0
14 h Base Address 1
18 h Base Address 2
1 ch Base Address 3
20 h Base Address 4
24 h Reserved
28 h Reserved
2 ch Subsystem ID Subsystem Vendor ID
30 h Expansion ROM Base Address
34 h Reserved Capabilities
Pointer
38 h Reserved
3 ch Max Latency Min Grant Interrupt Pin Interrupt
Line

10.1.1 PCI Vendor ID

Address offset 00h

Default Value ????h

Attribute Read Only

Size 16 bits

10.1.2 PCI Device ID

Address offset 02h

Default Value 1763h

Attribute Read Only

Size 16 bits

10.1.3 PCI Command Register

Address Offset 04h

Default Value 0000h

Attribute Read/Write

Size 16 bits

TABLE 3
PCI Command Register
Bit Description
 0 Target I/O enable. A value of 0 disables
Base Address Registers 0-4.
 1 Target ROM enable. A value of 0 disables
the Expansion ROM Base Address Register.
 2 Master enable. A value of 0 disables the Master
mode function of the controller.
 5-3 Reserved
 6 Parity check enable. A value of 0 causes parity
errors to be ignored
15-7 Reserved

10.1.4 PCI Status Register

Address Offset 06h

Default Value 0210h

Attribute Read/Clear

Size 16 bits

TABLE 4
PCI Status Register
Bit Description
 3-0 Reserved
 4 Capabilities Enable
 5 Set to one to indicate 66 MHz-Capable.
 7-6 Reserved
 8 Data Parity error detected
10-9 DEVSEL timing .01 b = Medium ??. These bits are read only
11 Signaled target abort
12 Received target abort
13 Received master abort
14 Reserved
15 Detected parity error

10.1.5 PCI Revision

Address Offset 08h

Default Value 00h

Attribute Read only

Size 8 bits

10.1.6 PCI Class Code

Address Offset 09h

Default Value 010180h

Attribute Read only

Size 24 bits

10.1.7 PCI Cache Line Size

Address Offset 0ch

Default Value 00h

Attribute Read/Write

Size 8 bits

10.1.8 PCI Latency Timer

Address Offset 0dh

Default Value 00h

Attribute Read/Write

Size 8 bits

10.1.9 PCI Header Type

Address Offset 0eh

Default Value 00h

Attribute Read only

Size 8 bits

10.1.10 PCI Base Address Registers (BAR)

All the Base Address Registers have Bit 0 hard-wired to 1 to indicate I/O space.

10.1.10.1 PCI Base Address 0

This is the base address for the command block registers for ATA channel X.

Address Offset 10h

Default Value 000001f1h

Attribute Bits 31-3 Read Write. Bits 2-0 Read only.

Size 32 bits

10.1.10.2 PCI Base Address 1

This is the base address for the control registers for ATA channel X. Note that, because of the d word alignment of PCI, the Device Control and Alternate Status Registers are at offset 02h from this base. For example, to put those registers at address 3f6h, this register must be set to 3f4h (+Bit 0).

Address Offset 14h

Default Value 000003f5h

Attribute Bits 31-03, Read Write. Bits 2-0, Read only.

Size 32 bits

10.1.10.3 PCI Base Address 2

This is the base address for the command block registers for ATA channel Y. Address Offset 18h

Default Value 00000171h

Attribute Bits 31-03, Read Write. Bits 2-0, Read only.

Size 32 bits

10.1.10.4 PCI Base Address 3

This is the base address for the control registers for ATA channel Y.

Address Offset 1ch

Default Value 00000375h

Attribute Bits 31-03, Read Write. Bits 2-0, Read only.

Size 32 bits

10.1.10.5 PCI Base Address 4

This is the base address for the ADMA registers.

Address Offset 20h

Default Value 00000001h

Attribute Bits 31-04, Read Write. Bits 3-0, Read only

Size 32 bits

10.1.11 PCI Subsystem Vendor ID

Address Offset 2ch

Default Value ????

Attribute Read only

Size 16 bits

10.1.12 PCI Sub-System ID

Address Offset 2eh

Default Value ????

Attribute Read Only

Size 16 bits

10.1.13 PCI Expansion ROM Base Address

Address Offset 30h

Default Value 000d0000h

Attribute Bits 31-16, Read Write. Bits 15-01, Read only. Bit 0, Read Write

Size 32 bits

10.1.14 PCI Capability Pointer

Address Offset 34h

Default Value 44h

Attribute Read Only

Size 8 bits

10.1.15 PCI Interrupt Line

Address Offset, 3ch

Default Value 00h

Attribute Read Write

Size 8 bits

10.1.16 PCI Interrupt pin

Address Offset 3dh

Default Value 01h

Attribute Read only

Size 8 bits

10.1.17 PCI Minimum Grant

Address Offset 3eh

Default Value 10h ?? (corresponds to 4 microseconds) ??

Attribute Read only

Size 8 bits

10.1.18 PCI Maximum Latency

Address Offset 3fh

Default Value ??h

Attribute Read only

Size 8 bits

10.1.19 PCI Timeout Registers

Address Offset 40,41h

Size 8 bits each

These are part of the Phoenix PCI core.

11 SFF Mode

11.1 ADMA Registers

The bus mastering function uses 16 bytes of I/O space (8 bytes per channel). All Host registers can be accessed as byte, word or d word entities. The description of these registers follows.

TABLE 5
ADMA Registers
Offset from
base address Register Access Mode
00 h Host Command-primary channel R/W
01 h Reserved
02 h Host Status-primary channel R/W/C
03 h Reserved
04 h-07 h PRD Table Address-primary channel R/W
08 h Host Command-secondary channel R/W
09 h Reserved
0 ah Host Status-secondary channel R/W/C
0 bh Reserved
0 ch-0 fh PRD Table Address-secondary channel R/W

11.2 ADMA Command Register.

Address offset, Primary channel: Base+00h

Address offset, Secondary channel: Base+08h

Default value 00h

Attribute Read/Write

Size 8 bits

TABLE 6
ADMA Command Register
Bit Description
7:4 Reserved; will return 0 on reads
3 R/W Control. This bit defines the direction of
the bus master data transfer
When 0, PCI reads (ATA writes) are performed
When 1, PCI writes (ATA reads) are performed.
This bit must not be changed while Bit 0 is 1.
2:1 Reserved-will return 0 on reads
0 Start Bit. This bit enables bus master transfers.
The transfer starts when this bit changes from 0 to
1. This bit shall be cleared to zero by the
Host software after the transfer is complete.
Transfers can be aborted by clearing this bit,
but data may be lost and the transfer
cannot be resumed.

11.3 ADMA Status Register.

Address offset, Primary Channel: Base+02h

Address offset, Secondary Channel: Base+0ah

Default value 00h

Attribute Read/Write/Clear

Size 8 bits

TABLE 7
ADMA Status Register
Bit Read Write Clear Description
7 X Cleared to zero to indicate concurrent channel
operation. (In simplex implementations, this bit
is set to one).
6 X Device 1, DMA capable. This bit reflects
the state of Bit 4.
5 X Device 0, DMA capable. This bit reflects
the sate of Bit 3.
4 X X Device 1. Ultra DMA capable. This bit
can be set by initializing software to
indicate that the drive is capable of
DMA transfers and has been initialized
for optimum performance
3 X X Device 0, Ultra DMA capable.
2 X X Interrupt-This bit is set by the rising
edge of the IDE interrupt line. This
bit can be cleared by writing a 1 to it.
When this bit is 1, all data has been
transferred successfully.
1 X X Error-This bit is set when the controller
encounters an error transferring
data. This bit can be cleared by writing
a 1 to it. Error details are in the
ATA Device Status Register or in the
PCI Status Register.
0 X Active-This bit is set when the Start Bit
of the Host Command Register
is set to 1.
This bit is cleared when the last PRD
is exhausted or the Start Bit is
cleared to 0. When this bit is 0, all data
has been transferred successfully,
unless the command was aborted.

11.4 ADMA PRD Table Address Register

Address offset, Primary Channel: Base+04h

Address offset, Secondary Channel: Base+0ch

Default value 00000000h

Attribute Read/Write

Size 32 bits

The PRD table must not cross a 64K boundary in memory.

TABLE 8
ADMA PRD Address Register
Bit Description
31:2 Base address of PRD table. Corresponds to D[31:2].
Must be dword-aligned.
 1:0 Set to zero

11.4.1 Physical Region Descriptor Table

Before the controller starts a Master mode transfer, it is given a pointer to a Physical Region Descriptor Table. This table contains one or more entries called Physical Region Descriptors (PRDs), which describe areas of memory that are involved in the transfer. The table must be aligned on a d word boundary and the table must not cross a 64K boundary in memory. The PRD Table must be contiguous in physical memory.

11.4.2 Physical Region Descriptor

Each Physical Region Descriptor (PRD) describes a contiguous region of memory involved in the transfer. The controller reads each PRD in turn and transfers data to or from the PRD-associated memory block, until the ATA Device interrupts to indicate the end of the transfer. If the current PRD space is exhausted, the controller reads the next PRD and continues.

If the last PRD space is exhausted and the ATA Device has not interruputed the controller indicates an error condition in its Status Register.

Every PRD is 8 bytes in length. Bytes 0 through 3 specify the 32-bit start address of a contiguous region of memory aligned to a d word boundary. Bytes 4 and 5 indicate the length of the region in bytes (a value of 0 indicates 64K). Byte 6 is reserved. Bit 7 of Byte 7 is cleared to zero for all PRD entries in the table, except the for the last entry, where Bit 7 is set to one.

TABLE 9
Physical Region Descriptor (PRD).
31 1 0
Memory Region Physical Base Address [31:2] 0 0
EOT Reserved [31:16] Byte Count [15:2] 0 0

11.5 Typical Programming Sequence

To initiate a bus master transfer between memory and an ATA Ultra DMA capable Device, the following steps are required.

1. Software prepares a PRD table in memory, starting on a dword boundary. Any number of 8 byte PRDs may be used, provided the table does not span a 64K boundary. The EOT Bit must be set in the last PRD. The total size of the PRD regions should be equal to or greater than the transfer size.

2. Software writes the start address of the PRD table into the Host PRD Table Address Register for the appropriate channel. Software defines the direction by writing Bit 3 of the Host Command Register and clears Bits 1 and 2 in Host Status Register.

3. Software writes the appropriate DMA command and parameters to the registers of the ATA Device.

4. Software starts the transfer by writing a 1 to Host Command Register, Bit 0, preserving the state of Bit 3.

5. The controller transfers data between the ATA Device and memory, responding to DMA requests from the Device.

6. At the end of the transfer, the ATA Device asserts its interrupt line. The controller sets Bit 2 and clears Bit 0 of the Host Status Register and asserts the PCI interrupt.

7. In response to the interrupt, software clears Bit 0 of the Host Command Register. It reads the Host Status Register and the ATA Device Status Register to determine if the transfer was successful.

11.6 Data Synchronization

When responding to the PCI interrupt or at the expiry of a time out period, the device driver must read the Host Status Register. If Bit 2 is set, then the ATA Device caused an interrupt, so all data transfer is complete. If Bit 2 is not set, then the ATA Device did not generate an interrupt, so the state of the transfer is unknown.

The controller guarantees that, when the PCI interrupt is generated in response to the ATA interrupt, all data transfer to or from the Device has been completed.

If the controller runs out of PRD space before the transfer is complete or encounters an error (such as a PCI target abort), it will not set Bit 2 of the Host Status Register or generate a PCI interrupt. In this case, the device driver shall be activated by a timeout.

11.7 Status bit Interpretation

The following table describes how to interpret Bit 2 and Bit 0 of the Host Status Register after a DMA transfer has been started.

TABLE 10
PRD Status bit Interpretation
Bit 2 Bit 0 Description
0 1 DMA transfer is in progress. No ATA
interrupt has been generated.
1 0 DMA transfer is complete. ATA interrupt
has been generated. Controller exhausted
its PRD table. This is the normal case.
1 1 DMA transfer is complete. ATA interrupt
has been generated. Controller did not
exhaust its PRD table. This is not an
error but the PRD table was larger
than necessary.
0 0 DMA transfer did not complete. This is
an error condition. If Bit 1 of the
Host Status Register is not set, the controller
ran out of PRD space. Otherwise, the reason
may be in the PCI Status Register
or the ATA Device Status Register.

If the DMA transfer did not complete, before attempting to recover from the error, the device driver must solve the problem of restoring the ATA Device to a state where it can respond to commands. The ATA Device may well be stuck with its DMARQ asserted, expecting more data to be transferred.

12 Auto DMA Mode

In Auto DMA Mode, the ADMA reads-in from host memory a command set held in the command chain. Command set are held in a data structure termed a Command Parameter Block (CPB). Control and status of the ADMA is maintained through the Chain Register. A chain of CPBs can be created in memory. This is a circular chain with each CPB pointing to the next CPB. The ADMA maintains a double-word register that points to the head of the chain. Within each CPB there is a pointer to the Physical Region Descriptor (PRD). The PRD is a structure that defines the memory locations where the data is to be written to or read from.

12.1 Command Parameter Block

The CPB is a block of parameters and commands for the ADMA and, indirectly, for the ATA channel. All transfers to and from the CPB are 32 bits wide. Any unused space is zero-filled.

13 Ultra DMA Timings

13.1 Ultra DMA Data Burst Timing Requirements

The following table includes a proposed timing for the Mode 5 operation (Ultra-100)

TABLE 11
Ultra DMA Data Burst Timing Requirements
Mode 0 Mode 1 Mode 2 Mode 3 Mode 4 Mode 5
Signal (ns) (ns) (ns) (ns) (ns) (ns) Comment
Name Min Max Min max min max min max min Max Min Max (see Notes 1 and 2)
T2CYCTYP 240 160 120 90 60 40 Typical sustained average two-cycle time
TCYC 112 73 54 39 25 18 Cycle time allowing for asymmetry and clock
variations (from STROBE edge to STROBE edge)
T2CYC 230 154 115 86 57 39 Two-cycle time allowing for clock variations (from
rising edge to next rising edge or from falling edge
to next falling edge - of STROBE)
TDS 15 10 7 7 5 3 Data setup time at recipient (see note 4)
tDH 5 5 5 5 5 1 Data hold time at recipient (see note 4)
tDVS 70 48 30 20 6 5 Data valid setup time at sender (from data valid
until STROBE edge) (see Note 5)
tDVH 6 6 6 6 6 2 Data valid hold time at sender (from STROBE
edge until data may become invalid) (see Note 5)
tFS 0 230 0 200 0 170 0 130 0 120 0 120 First STROBE time (for device to first negate
DSTROBE from STOP during a data in burst)
tLI 0 150 0 150 0 150 0 100 0 100 0 100 Limited interlock time (see Note 3)
tMLI 20 20 20 20 20 20 Interlock time with minimum (see Note 3)
tUI 0 0 0 0 0 0 Unlimited interlock time (see Note 3)
tAZ 10 10 10 10 10 10 Maximum time allowed for output drivers to
release (from asserted or negated)
tZAH 20 20 20 20 20 20 Minimum delay time required for output
tZAD 0 0 0 0 0 0 Drivers to assert or negate (from released)
tENV 20 70 20 70 20 70 20 55 20 55 20 35 Envelope time (from DMACK- to STOP and
HDMARDY- during data in burst initiation and
from DMACK to STOP during data out burst
initiation)
tSR 50 30 20 NA NA NA STROBE-to-DMARDY- time (if DMARDY-
is negated before
this long after STROBE edge recipient
shall receive no more than 1 additional data word)
tRFS 75 70 60 60 60 60 Ready-to-final-STROBE time (no STROBE edges
shall be sent this long after negation of DMARDY-)
tRP 160 125 100 100 100 100 Ready-to-pause time (that recipient shall wait to
pause after negating DMARDY-)
tIORDYZ 20 20 20 20 20 20 Maximum time before releasing IORDY
tZIORDY 0 0 0 0 0 0 Minimum time before driving IORDY (see note 6)
tACK 20 20 20 20 20 20 Setup and hold times for DMACK- (before
assertion or negation)
tSS 50 50 50 50 50 50 Time from STROBE edge to negation of DMARQ
or assertion of STOP (when sender stops a burst)
NOTES -
1 Timing parameters shall be measured at the connector of the sender or receiver to which the parameter applies. For example, the sender shall stop generating STROBE edges tRFS after the negation of DMARDY-. Both STROBE and DMARDY- timing measurements are taken at the connector of the sender
2 All timing measurement switching points low to high and high to low) shall be taken at 1.5 V
3 tUI, tMLI and tLI indicate sender-to-recipient or recipient-to-sender interlocks, i.e., one agent (either sender or recipient) is waiting for the other agent to respond with a signal before proceeding. TUI is an unlimited interlock that has no maximum time value. tMLI is a limited time-out that has a defined minimum tLI is a limited time-out that has a defined maximum.
4 Special cabling shall be required in order to meet data setup (tDS) and data hold (tDH) times in Modes 3 and 4
5 Timing for tDVS and tDVH shall be met for all capacitive loads from 15-40 pf. where all signals have the same capacitive load value.
6 tZIORDY has no maximum and may be greater than tENV, since the device has a pull-up on IORDY-, giving it a known state when released

TABLE 12
CPB Structure
Double
Word Byte Bits Value/Range Description
0 Address Fields
0 0-7 0-255 ADMA Sub-channel Number
1 0 0/1 Bit 0 = ATA Device Number on
the channel
1 0/1 Bit 1 = Cleared to zero indicates
non-queued command; set to
one indicates queued command
sequence.
2 0 Bit 2 = Reserved; shall be set
to zero.
3-7 0/1 Bit 3-7 = ATA Tag if Bit 1
set to one; else reserved.
2 0-2 0-4 Bit 0-2 Command Block PIO
Mode (Modes 0-4)
3-7 0 h Reserved
3 0-1 0-3 Device Data Transfer method
2-3 016 Device Data Transfer Mode
4 0/1 Shall be cleared to zero to
indicate that Double Word 3
contains a PRD Address.
Shall be set to one to
indicate that Double Word 3
contains an I/O address for
the transfer.
3-7 0 Reserved
1 0-3 4 Valid Memory Memory Address of the
Address Next CPB.
2 Command and Information Fields
0 0-7 Variable ADMA Command Byte
(see description)
1 0-7 0 h-FFh CPB ID
2-3 0-15 Variable ADMA Status Word
3 0-7 Variable ADMA Chain Byte
3 0-3 4 Valid Memory PRD or I/O Address
Address
4 0-1 2 Variable ATA Command Data-See
description
2-3 2 Variable ATA Command Data-See
description
. . . . .
. . . . .
. . . . .
n 0-1 2 Variable ATA Command Data-See
description
2-3 2 Variable ATA Command Data-See
description

13.1.1 Address Fields (Double Word 0)

13.1.1.1 ADMA Sub-channel Number (Byte 0)

00h indicates the base channel, 01h-FFh are valid sub-channel addresses. 00h is the power on and reset default value.

Sub-channel addresses are changes using channel commands. These commands are snooped by the ADMA channel command unit. The currently selected sub-channel number is maintained by that unit. A CPB that contains a sub-channel command changing the current sub-channel will have the ADMA sub-channels number set to the current sub-channel. The next CPB will address the new sub-channel number.

13.1.1.2 ATA Device ID (Byte 1, Bit 0)

This bit cleared to zero indicates that the command sequence is for device zero on the currently selected channel. This bit set to one indicates that Device 1 is the selected device on the currently selected Channel.

The command sequence must start with a write to the ATA Device Head Register. The Device Bit in this write and any other Device Head Register write within the command sequence of a CPB must correspond to the setting of the ADMA Device ID Bit. I. If the bits do not correspond the CPB will be aborted and the ADMA will halt with an error status.

13.1.1.3 Overlap/Queue Flag (Byte 1, Bit 1)

This bit is set to one to indicate that the command set contains an Overlapped command. This bit being cleared to zero indicates that there is no queued command. If this flag is set the ADMA will inspect the SERV Bit in the ATA Status Register and the REL Bit in the ATA Sector Count Register on the assertion of ATAINTRQ.

13.1.1.4 ATA Tag (Byte 1, Bits 3-7)

If Bit 1(13.1.1.3) is to one this field contains the ATA TAG number associated with the overlapped command.

13.1.2 Channel Performance Parameters (Bytes 2,3)

The performance parameters for the device channel. These values will stick in the ADMA, on a operating mode change.

13.1.2.1 ATA Device Register PIO Mode (Byte 2, Bits 0-2)

The PIO Mode, as defined in ATA-5, to use when performing Command Block register transfers. For two Devices on a channel, this should be set to the lower of the two mode values reported by the Devices.

13.1.2.2 ATA Device-n Data Transfer Method (Byte 3, Bits 0-1)

00h=PIO

02h=Ultra DMA

All other values=Reserved.

13.1.2.3 ATA Device-n Data Transfer Mode (Byte 3, Bits 2-3)

If in PIO mode the operating mode defined in ATA-5. If in Ultra DMA mode the mode number as defined in the timing table XX.

13.1.3 ADMA Command Byte

This Byte is written by the Host to control the execution of the command set. When the CPB is loaded the contents are mapped into the same I/O space as the ADMA Command Register. The ADMA Command Register may be read but not written in Auto DMA Mode. The meanings on some of the bits change when in this mode. The contents of this Byte are not changed by the ADMA when the ADMA status is updated. Thus the host software has a record of the initial command conditions of the CPB.

The contents of this Byte and the ADMA Status Byte are examined during the ADMA scan phase. To ensure the CPB Ready Bit in the Command Byte must be set to one and the CPB Complete Bit in the ADMA Status Byte must be set to zero before the CPB is valid and ready for execution.

TABLE 13
ADMA Command Byte and Register
Bit Description
7:3 Reserved; shall be cleared to zero.
4 pINTA inhibit bit is set to one if the
ADMA is NOT to interrupt the host when
the CPB has completed.
3 R/W Control. This bit defines the direction
of the bus master data transfer
When 0 PCI reads (ATA writes-data transferred
to the ATA Device) are performed
When 1, PCI writes (ATA reads-data transferred
from the ATA Device) are performed.
2 Reserved-will return 0 on reads
1 ATAP bit is set to one to indicate that the
command set is for an ATAPI command and the ATA
Byte Count Registers are significant.
0 CPB Ready Bit. This bit shall be set cleared
to zero before updating ANY of the control
fields of a CPB. When the CPB is ready to be
executed this bit shall be set to one.

13.1.4 CPB ID

The Host inserts an identification number from 0h to FFh in this field. The value is loaded into the ID register by the ADMA. The host may read this register to determine which CPB is being executed. The ADMA takes no action on this value, it is the host's responsibility to ensure that the value is unique within the chain. Note that for practical reasons the CPB should be unique, since the CPB ID has no operational function in the ADMA the values can be duplicated and thus the chain can be of infinite length.

13.1.5 ADMA Status Byte

This Byte shall be cleared to 00h by the Host when initializing the CPB. On return the ADMA will load the Byte the contents of the ADMA Status Register at the termination of the appropriate ADMA phase. The ADMA Status Register is read only when the ADMA is in Auto DMA mode. The bit settings in this Byte (Register) shall different meanings to those in SFF mode.

TABLE 14
ADMA Status Byte and Register
Bit Description
7 This bit is cleared to zero by the Host when loading
the CPB entries. This bit is set to one by the UDMA
when the CPB has been completed. This bit is not set
when a overlapped command has released the bus.
6 This bit is set to one by the UDMA when a Overlapped
command has released the bus. This bit is cleared
by the UDMA when the Service command has been
issued.
5 Set to one to indicate that the ADMA detected
a CPB entry error.
4 PCI Error-The UDMA detected an error in the PCI
transfer on the PCI bus. The error code will be
in the PCI Status Register.
3 Forced Error-This bit is set when the ADMA forced
a CRC Error in Ultra DMA mode.
2 The ATA Device Asserted ATA-INTRQ
1 ATA Error-Indicates that the ATA-ERR (CHK)
Bit is set in the ATA status register. This bit
should be set if Bit 3 is set
0 Active-This bit is set to one by the UDMA when
the Data transfer is in progress. This bit is cleared
to zero when the last PRD is exhausted.

13.2 ADMA Chain Byte

13.2.1 Chain Register

TABLE 15
ADMA Chain Byte and Register
0 1 2 3 4 5 6 7
ACTV PSCHN PSCPB NTFY CINTRQ UINTRQ CPBFN 0
R/W R/W R/W R/W R R R R

13.2.1.1 ACTV-CPB Active (Bit 0)

The Host sets this bit to one whenever it adds a CPB to the CPB chain or changes the status of a CPB to “Service Required”. The ADMA clears this bit to zero whenever it accesses the CPB chain in search mode.

13.2.1.2 PSCHN—Pause Chain (Bit 1)

The Host sets this bit to one to pause any further processing of the CPB chain. The Host shall clear this bit to zero to allow the ADMA to scan and process the CPB chain.

13.2.1.3 PSCPB—Pause the Current Active PCB (Bit 2)

If the Host wishes to terminate a currently active PCB, it will set this bit to one. The ADMA will clear this bit if the CPB cannot be paused.

13.2.1.4 NTFY Notify When Inactive

If set to one the ADMA will interrupt the host when there are no more entries in the CPB chain that it can currently service.

13.2.1.5 CINTRQ Channel Interrupt

The INTRQ line on this channel caused a pINTA (PCI interrupt) to occur. It should be noted that this may not be the only reason a PCI interrupt has occurred.

13.2.1.6 UINTRQ Channel Interrupt

The UINTRQ line (unsolicited interrupt) on this channel caused a pINTA (PCI interrupt) to occur. It should be noted that this may not be the only reason a PCI interrupt has occurred.

13.2.1.7 CPBFN CPB Chain Finished

There are no more CPB entries that are either not started or released. If NTFY is set the ADMA will have interrupted.

13.3 ATA Command Data

Bits 0-2 are reserved and set to zero. Bits 3-7 are used to indicate the ATA command or control register to address. The bits directly map onto the register address to be mapped. Bits 8-15 contains the bit pattern to be written to the register.

TABLE 16
ATA Command Data
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
0 0 0 CS0 CS1 AD0 AD1 AD2 0/1 0/1 0/1 0/1 0/1 0/1 0/1 0/1

The ADMA scans in the CPB reading the ATA Command data double words into its FIFO. When the ADMA detects the address decode for the ATA Command Register it will stop scanning. The maximum number of ATA Command Data double words is 12 (max CPB size is 16 in total). If this size is exceeded before an ATA Command Register Address is encountered the ADAM will halt with an error condition.

14 CPB Execution

14.1 Command Parameter Block Chain

Whenever the Host adds an entry to the chain, it will set ACTV to one. The ADMA will search along the chain, looking for CPBs that need service. If no CPBs need service, the ADMA will cease to search the chain, thereby reducing PCI and memory bus utilization.

14.2 CPB Execution Phases

14.2.1 Parameter Scan Phase

This phase is entered from the ATA Bus Release Phase, the Command Completion Phase or the host writing a Start Bit into the ADMA command Register. The ADMA searches the CPB chain for the next startable CPB. If none is found, the ADMA goes to its idle state.

14.2.2 Parameter Delivery Phase

Once the ADMA has determined that it has a valid executable CPB, it will enter the Parameter Delivery Phase. This phase is completed when all the register contents have been dispatched to the ATA channel, except the contents of the ATA Command Register. If the ADMA encounters an invalid register address or a register read indicator before the ATA Command Register, it will indicate an error and enter the Command Completion Phase.

14.2.3 Command Initialization Phase

The request to write the ATA Command Register indicates that all the parameters have been loaded and the Command Initialization Phase has been entered. This phase exists until the ATA Device indicates that the command has been accepted (either ATA-BSY is set and then cleared or the Device interrupts and ATA-BSY is cleared). If the Device indicates an error, the ADMA then enters the Command Completion Phase. If a queued command has been accepted, it will either be serviced immediately and be treated like a regular command or it will indicate a bus release.

14.2.4 ATA Bus Released Phase

If the command is a queued command and the bus has been released by the ATA Device, the ADMA enters the Bus Released Phase.

14.2.5 Tag Search Phase

This phase is entered when a Device interrupts with no command in execution. The ADMA reads the ATA Status Register and checks the SERV Bit. This will be set to one to indicate service is required. The ADMA will then issue a Service command to the ATA Device. The ATA Device will interrupt and the ADMA will read the ATA sector count register to obtain the TAG number. The ADMA then reads through the CPB chain looking for a released CPB that has a matching sub-channel, Device and Tag ID. It then reads in the ADMA configuration data and completes the transfer. If no CPB is found, the ATA Device indicates an error, or the SERV Bit was not set, the ADMA enters the Command Completion Phase, indicating an error.

14.2.6 Data Transfer Phase

If the ATA command involves data transfer and it is not a released queued command, the Data Transfer Phase is entered. This phase is continued until the transfer reaches normal completion or is prematurely terminated by the ATA Device or by a request from the Host. At the end of the transfer the ADMA reads the ATA Status Register and examines the Error Bit.

14.2.7 Command Completion Phase

This phase can be entered from any phase. The ADMA will update the ADMA status register Byte in the current CPB. PCI INTAn is asserted. If there are no errors, the Parameter Scan Phase is entered; otherwise, the ADMA halts.

14.3 Host Pausing of a CPB Chain

The Host may attempt to pause an active chain, at any time, by setting PSCHN to one. The ADMA will complete any currently active CPB but will not initiate a search of the CPB chain. Upon completion of the current CPB or if no CPB is being serviced, the ADMA will assert INT-HST. The Host may then modify the CPB-SW of any of the CPB entries in the chain.

14.4 Host Termination of an Active CPB

Once the Host has written a CPB, it will normally not update the CPB, until the ADMA has indicated that the command is completed. The CPB may then be reused for another command set. In some circumstances, the Host may wish to terminate a command set before total completion.

To do this, the Host will attempt to pause the ADMA by setting the Pause-CPB Bit in the Chain Register. The Host then reads the Chain Register. If the current CPB can be paused, the Pause-PCB Bit will be set. If the ADMA cannot pause the current CPB, the Pause-CPB Bit will be cleared. If the Pause-CPB Bit is set, the Host shall then read the Current CPB Register. If the current CPB is the CPB which it wishes to terminate, the Host shall then write the Terminate Bit in the Chain Register. If the current CPB is not the target CPB, the Host may update the CPB-SW, to indicate that it does not need service. After either action, the Host shall clear the Pause-CPB Bit.

The Host will then wait for normal completion of the CPB and examine the CPB-SW to determine how far the CPB had been processed prior to termination.

14.4.1 ADMA's Response to a CPB Termination Request.

To terminate a currently active CPB, the Host must follow the protocol described in Section 14.4. If the Host attempts to set the Terminate Bit in the Chain Register without the Pause Bit being set, the Termination Bit shall be ignored by the ADMA.

14.4.1.1 Already Paused

If the Termination Bit has to be set while the ADMA is paused, the action taken will depend on the CPB's execution phase. While in the Parameter Delivery Phase, the ADMA will immediately transition to the Command Completion Phase, indicating the reason in the CPB-SW.

14.4.1.2 Dat Transfer Phase

If in the Data Transfer Phase, the ADMA will attempt to terminate the transfer. In ADMA Mode, it will do this by terminating the ADMA-Burst and deliberately corrupting the CRC. This may result in the ATA Device prematurely terminating the command. This action will result in an error indication in the ATA Status Register. Upon transition to the Command Completion Phase, the reason for the termination will be indicated and the actual transfer length will be indicated.

14.4.1.3 Initialization Phase

During the Command Initialization Phase, the CPB cannot be terminated. If the subsequent phase is the Data Transfer Phase, the ADMA will respond as described in 14.4.1.2.

15 Channel Command Interpreter

The channel command interpreter monitors all ATA command traffic. It will respond to requests for ADMA identification and revision information, auto polling implementation, unsolicited interrupt and HotPort™ change notification.

The unit also keeps track of the current sub-channel number to enable CPB ID matching.

Patent Citations
Cited PatentFiling datePublication dateApplicantTitle
US5129064 *Aug 27, 1990Jul 7, 1992International Business Machines CorporationSystem and method for simulating the I/O of a processing system
US5404550 *Jul 25, 1991Apr 4, 1995Tandem Computers IncorporatedMethod and apparatus for executing tasks by following a linked list of memory packets
US5561816 *Jan 11, 1993Oct 1, 1996Nec CorporationData transfer controlling device for use in a direct memory access (DMA) system
US5721828 *Nov 5, 1996Feb 24, 1998Mercury Computer Systems, Inc.Multicomputer memory access architecture
US5860083 *Mar 14, 1997Jan 12, 1999Kabushiki Kaisha ToshibaData storage system having flash memory and disk drive
US5903281 *Mar 7, 1996May 11, 1999Powertv, Inc.List controlled video operations
US6161154 *Nov 26, 1997Dec 12, 2000National Instruments CorporationSystem and method for extracting and restoring a video buffer from/to a video acquisition cycle
Referenced by
Citing PatentFiling datePublication dateApplicantTitle
US6772311 *Jun 24, 2002Aug 3, 2004Texas Instruments IncorporatedATAPI device unaligned and aligned parallel I/O data transfer controller
US6880033 *Apr 17, 2002Apr 12, 2005Adaptec, Inc.Methods for configuring separate accessibility of each channel of a dual channel SCSI chip
US7058748 *Jun 28, 2002Jun 6, 2006Cypress Semiconductor Corp.ATA device control via a packet-based interface
US7130932 *Jul 8, 2002Oct 31, 2006Adaptec, Inc.Method and apparatus for increasing the performance of communications between a host processor and a SATA or ATA device
US7130933 *Jul 24, 2002Oct 31, 2006Intel CorporationMethod, system, and program for handling input/output commands
US7366872 *Dec 30, 2003Apr 29, 2008Intel CorporationMethod for addressing configuration registers by scanning for a structure in configuration space and adding a known offset
US7516282 *Sep 8, 2006Apr 7, 2009Ite Tech. Inc.Control device and control method for memory
US7610444 *Oct 6, 2006Oct 27, 2009Agere Systems Inc.Method and apparatus for disk address and transfer size management
US7822893May 17, 2007Oct 26, 2010Samsung Electronics Co., Ltd.System and method for data transfer using ATA interface
US7827320Mar 28, 2008Nov 2, 2010Western Digital Technologies, Inc.Serial ATA device implementing intra-command processing by detecting XRDY primitive while in the XRDY state
US7929533 *Mar 24, 2003Apr 19, 2011British Telecommunications PlcSystem for selecting a connectivity mechanism
US8364863 *Dec 18, 2008Jan 29, 2013Intel CorporationMethod and apparatus for universal serial bus (USB) command queuing
US8473798 *Mar 22, 2010Jun 25, 2013Comtect EF Data Corp.Encoding and decoding systems and related methods
US8856390Mar 28, 2008Oct 7, 2014Western Digital Technologies, Inc.Using device control field to implement non-disruptive notification of an ATA device
US8930583Sep 18, 2003Jan 6, 2015Marvell Israel (M.I.S.L) Ltd.Method and apparatus for controlling data transfer in a serial-ATA system
Classifications
U.S. Classification710/22, 710/43, 710/24, 710/48, 710/36, 710/39
International ClassificationG06F13/28, G06F13/20
Cooperative ClassificationG06F13/28
European ClassificationG06F13/28
Legal Events
DateCodeEventDescription
Apr 15, 2008FPExpired due to failure to pay maintenance fee
Effective date: 20080224
Feb 24, 2008LAPSLapse for failure to pay maintenance fees
Sep 3, 2007REMIMaintenance fee reminder mailed
Sep 29, 2003ASAssignment
Owner name: COMERICA BANK-CALIFORNIA, CALIFORNIA
Free format text: SECURITY INTEREST;ASSIGNOR:PACIFIC DIGITAL CORPORATION;REEL/FRAME:014540/0343
Effective date: 20020110
Owner name: COMERICA BANK-CALIFORNIA 2052 ALTON PARKWAYIRVINE,
Owner name: COMERICA BANK-CALIFORNIA 2052 ALTON PARKWAYIRVINE,
Free format text: SECURITY INTEREST;ASSIGNOR:PACIFIC DIGITAL CORPORATION;REEL/FRAME:014540/0343
Effective date: 20020110
Feb 21, 2001ASAssignment
Owner name: PACIFIC DIGITAL CORPORATION, CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:GOODFELLOW, TONY;REEL/FRAME:011541/0907
Effective date: 20010212
Owner name: PACIFIC DIGITAL CORPORATION 2052 ALTONIRVINE, CALI
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:GOODFELLOW, TONY /AR;REEL/FRAME:011541/0907
Owner name: PACIFIC DIGITAL CORPORATION 2052 ALTONIRVINE, CALI
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:GOODFELLOW, TONY /AR;REEL/FRAME:011541/0907
Effective date: 20010212